Significant deficits in language and communication are among the defining features of Autism. Special educators working with children presenting Autism urgently require effective, efficient, and practical techniques for teaching basic linguistic skills and the symbolic relationships between stimuli that provide the supporting framework for such skills. Applied behavior analysis has proven to be an effective evidence-based approach for this task and discrete-trial training techniques have featured as a mainstay of this approach. Matching-to- sample (MTS) tasks are a subset of these techniques and are often used to teach symbolic relationships between words and their referents. Various methods for teaching MTS to children with Autism have been described in the research literature. Most methods involve either manipulating the stimuli throughout training or extrinsic prompting techniques. However, for various reasons, they are seldom used in applied settings. Instead, MTS tasks are most often taught by a teacher using flashcards in table-top methods that involve some degree of trial-and-error learning. Unfortunately, only a few published studies have examined how the efficacy of this basic method can be enhanced by manipulating the consequences arranged for correct responses and errors, but these studies suggest significant potential for improvement. The research proposed here addresses this knowledge gap. It involves a component analysis of the error- and preference-correction procedures that have supplemented trial-and-error training on MTS tasks with a view to identifying the behavioral processes underlying their effectiveness. Three laboratory studies will each use children with Autism and compare their rates of learning the intended stimulus associations across MTS procedures that differ with respect to an isolated and exaggerated feature of some correction technique. The independent measures of conditional stimulus-control and response bias offered in recent quantitative models of MTS performance will be used to monitor changes in participants'performance with increasing exposure to the procedures, and the procedures will be arranged in an alternating-treatments design. These experiments will determine: (1) whether error-correction trials serve a punitive or a practice function by comparing the efficacy of immediate correction with delayed correction, (2) whether the remedial effects of the standard error-correction procedure accrue from preventing the development of stimulus or position biases by controlling relevant distributions of reinforcers, (3) whether explicitly increasing the opportunities for errors to extinguish, and/or increasing stimulus observing requirements, on error-correction trials enhances their efficacy. A fourth study, conducted in a classroom using special-education teachers and educationally-relevant stimuli, will assess the social validity of an effective correction procedure from the lab studies. An anticipated outcome of this research is the design and empirical validation of procedures that can either stand alone or supplement others (including those aiming to train in an "errorless" manner) to speed the learning of MTS tasks in people with moderate to severe autism. PUBLIC HEALTH RELEVANCE: The proposed research is highly relevant to the public health interests of NICHD and its Mental Retardation Developmental Disabilities branch because it will contribute directly to the improvement of teaching methods for individuals with autism and other moderate to severe intellectual disabilities. Specific areas of application include improvements in functional communication capabilities and fundamental linguistic skills for children with moderate to severe disabilities.